Multi-Factor-Asset Pricing Models for German Stocks: An Empirical Analysis of Time Varying Parameters
Posted: 10 May 2004
Date Written: January 2004
The large number of asset pricing models and empirical studies of stock returns are evidence of the desire to understand the return generating process of financial assets in general and for stocks in particular. One focus of the research in this area has been on multi-factor asset pricing models [Chen et al. (1986), Fama/French (1992)]. These models are based on the assumption that stock returns are generated by a limited number of economic variables such as company, industry or macroeconomic factors.
The objective of this study is to analyze the importance of various economic factors in explaining the return structure for stocks in Germany and to investigate whether the impact of these factors is time varying. This is important, because in most studies of asset pricing models it is assumed that the parameters are non time varying. In particular, we investigate the time variability of the explanatory power and the beta coefficients in a multi-factor framework. For this we employ a rolling estimation procedure that allows us to analyze the time variability of the model coefficients.
In the empirical analysis we use monthly data of four macroeconomic variables and the market index to explain the returns of four German industry indices for the period from 1974 to 2000. In contrast to most studies which exclude banks from their empirical analysis we use three industrial indices and a bank index. The economic factors included in our model are term spreads, interest rates, exchange rates and the ifo business index as well as the market index. The empirical results confirm that the factors used in our empirical analysis seem well suited to explain the stock returns especially for banks. Moreover, it is evident that the explanatory power and the beta coefficients are time varying.
Suggested Citation: Suggested Citation